Synechococcus elongatus recoding-diversity map

How to read this

What a dot is

One dot is one gene. There are about 2,700 of them. Wherever you see a scatter of dots on this page, every dot is the same gene it was on the previous plot, so a gene you pin stays highlighted when you switch views.

What “close together” means

The PCA and UMAP maps combine many gene measurements into two dimensions; nearby genes are similar in the inputs to that map, and those axis numbers have no native units. Metric X vs Y instead plots the two measurements you choose, in their own units.

Under a PCA map, What drives these axes lists measurements that pull genes left, right, up, or down. The metric plot names its two measurements directly.

Why these maps differ

They answer different questions. Native codon space asks how a gene writes its codons today. Recoding-risk space asks how exposed a gene is to the scheme you set. Baseline risk UMAP is a grouping view: it clusters similar genes tightly, but the gaps between clusters mean nothing. Perturbation space asks how far recoding would move a gene from where it started, which is a different question from where it started. Metric X vs Y shows a direct relationship, such as CDS length against CAI. A recoded genome may also offer a separately labelled public-parent frame. That map keeps the recoded genome's native refit available and does not turn sequence separation into a fitness or expression measurement.

Setting a recoding scheme

A scheme is a list of codons you want to remove, each paired with the synonymous codon that replaces it. The replacement is prefilled with the codon this genome already uses most for that amino acid, and you can change it to any synonymous alternative. You cannot pick a replacement that changes the protein; the page checks every gene's real sequence and says so underneath.

What counts as an edit

Burden numbers count the codons inside the gene that the scheme would replace, plus the gene's final stop codon when the scheme reassigns stops. Two codons are deliberately left out. The first codon is never recoded: it is read as methionine whatever the triplet says, so swapping it would change the protein even though the codon table calls it synonymous. The final stop is counted as an edit but is not part of any density window or cluster, which are measured across the coding codons only.

Genes that need a warning

A few genes only translate correctly because the ribosome does something unusual, such as shifting frame partway through. Those are flagged in the gene panel and can be filtered on. They are poor recoding candidates, so the page shows the flag rather than quietly dropping them.

Reading a percentile

In the gene panel, “78th pct” means this gene's value is higher than 78 % of genes in the genome. That is usually more useful than the raw number, because it tells you whether the gene is ordinary or unusual.

Reading a change

With a scheme set, each gene gets a wild-type value, a recoded value, and the difference. Both values are recalculated here in the same way, so the difference is fair. A negative ΔCAI or ΔtAI means the recoded gene is a worse match for this organism's translation machinery, which is usually the thing you are trying to avoid.

Where numbers come from

Most numbers here are computed from this genome's own sequence. Expression is the exception: when a dataset carries it, it may have been measured in a different strain or condition, so wherever it appears the page prints its source next to it. It is never the default way to judge whether a gene is worth keeping; codon adaptation and tRNA adaptation are, because those come from this genome. A gene with no measurement is unknown, not zero, and is kept unless you choose otherwise.

Colour

Colour is never the only signal. Shortlisted genes also get a diamond outline, the pinned gene gets a ring with crosshairs, an active gene reached by arrow key gets a third, distinct ring while you decide whether to pin it, and genes hidden by a filter shrink to small grey dots.

Moving the map and sharing a link

Moving around the map

Drag to pan and scroll inside the grid to zoom with a mouse or trackpad; the Zoom in and Zoom out buttons do the same on a touch screen, with no pinch gesture required. With the map focused, arrow keys move an active gene and announce it without pinning anything, Enter pins that active gene, and S adds or removes it — or, if nothing is active yet, the pinned gene — from the shortlist.

Sharing a link

The address bar always matches what is on screen, including while you paste in a different link or use the browser's back and forward buttons: the page applies the new view live, without a reload. A link encodes its shortlist explicitly, even an empty one, so sending someone a cleared shortlist actually clears theirs instead of quietly showing whatever they had saved locally. Anything a link does not mention — most often, on a first visit with no link at all — falls back to what this browser last saved, and only then to the page's defaults.

By Rohin Shanker

Loading genes

Controls

Drag to pan, scroll inside the grid or use the Zoom in/out buttons to zoom, double-click or 0 to reset. Scroll over the axes or margins to scroll the page.
With the map focused: arrow keys move an active gene without pinning it and announce what it is; Shift and an arrow pans; plus and minus zoom.
Enter pins the active gene. S adds or removes the active gene (or, with nothing active, the pinned gene) from the shortlist. Click a pinned dot again to unpin it.
The zoom buttons also work by touch.

What drives these axes

Candidate shortlist

How the shortlist connects to comparison

Add candidates from search, the map, or gene detail. The comparison above updates immediately; choose a row here to reopen its gene detail.

About this dataset

View provenance and metric conventions